The eNTERFACE'05 EMOTION Database

The eNTERFACE'05 EMOTION Database is an audio-visual emotion database that can be used as a reference database for testing and evaluating video, audio or joint audio-visual emotion recognition algorithms. Additional uses may include the evaluation of algorithms performing other multimodal signal processing tasks, such as multimodal person identification or audio-visual speech recognition. It was designed during the eNTERFACE'05 workshop, in the framework of a project on multimodal caricature [1,2].

CONTENTS

The final version of the database thus contains 42 subjects, coming from 14 different nationalities. Among the 42 subjects, a percentage of 81% were men, while the remaining 19% were women. A percentage of 31% of the total set wore glasses, while 17% of the subjects had a beard.
The recordings lasted for two weeks. All the experiments were driven in English. Each subject was told to listen to six successive short stories, each of them eliciting a particular emotion. They had then to react to each of the situations and two human experts judged whether the reaction expressed the emotion in an unambiguous way. If this was the case, the sample was added to the database. If not, it was discarded.

TECHNICAL ASPECTS

The database was recorded using a standard mini-DV digital video camera. The resolution of the camera was 800.000 pixels. The recording of the speech signal was realized through the use of a high-quality microphone, specially conceived for speech recordings. The microphone was situated roughly 30cm below the subject's mouth, outside of the camera field.
The background consists of a monochromatic dark gray panel that covered the entire area behind the subject, to allow easier face detection and tracking. Illumination was made constant through the use of a set of occultation panels, placed in front of every window. Lighting material consisted of a strong spotlight (500 watts), situated right behind the camera, facing the user. The spotlight was covered with a semi-transparent plastic film to soften the light, decrease the shadows and protect the subject from the very intense source of light. Two additional directional spots were situated between the subject and the background panel, so as to cancel shadows produced on the background panel by the main spotlight. The two additional spots were also covered with a semi-transparent plastic film.
The doors remained closed at all time to prevent external sound to interfere with the experiments.

LICENSE

This database is available under MIT license conditions (the terms of this very open license are provided with the database).

DOWNLOAD

Please note that the database is very big (0.8 GB). Click HERE to proceed.

[1] : Martin, O., et al., "Multimodal Caricatural Mirror", in in Proc. eNTERFACE 2005, July 18th-August 12th, Mons, Belgium - available on the eNTERFACE '05 website : http://www.tcts.fpms.ac.be/enterface//enterface05/docs/results/reports/project2.pdf.
[2] : O. Martin, I. Kotsia, B. Macq and I. Pitas : "The eNTERFACE'05 Audio-Visual Emotion Database", Proceedings of the First IEEE Workshop on Multimedia Database Management, Atlanta, April 2006.